On Mon, May 07, 2001 at 12:24:33AM -0500, Richard Rauch wrote:
> 
> _Are_ the VIA chipset problems exacerbated under 1.5.1?  Or, can I
> reasonably expect to remain as pleased with my VIA/Athlon system under
> 1.5.1 as I am under 1.5?  It seems that the displeasure with VIA has
> mounted of late, but my hazy impression is that the people doing the
> complaining are more prone to running -current or the 1.5.1_ALPHA or _BETA
> releases.  On the other hand, no one speaks of reverting to 1.5 to solve
> their problems...

Let me toss one more data point onto the graph.

On (well, under :-) my desk at work is a 1.3GHz K7 box that uses the
AMD 761 north bridge and the VIA 686B south bridge.  I have NOT turned
off any of the chipset features (PCI delayed transaction, bursting,
etc.) that allegedly must be turned off to avoid trouble.  I have four
UDMA/100 disks on this box (all identical); one on each channel of the
686B and one on each channel of a Promise UDMA/100 controller.

I can (and do!) sustain almost 90MB/sec to the disks.  I've rebuilt
NetBSD repeatedly, built much of an XFree 4 tree, built all kinds of
stuff for work, and built much of pkgsrc (KDE, koffice, gnome, and
everything needed for those, as well as most of pkgsrc/x11).

I have no idea what's wrong with other people's 686B-based systems
(and I note that the first reports of VIA chipset problems were all
from people with VIA KX133 chipset motherboards, which certainly did
NOT have the 686B south bridge on them!) but this one has had no
trouble at all.  No data corruption, no metadata corruption, no
random crashes -- no problems to report, period.

This leaves me wondering if the actual bug that's corrupting data
here is not in the IDE controller at all, but perhaps in VIA's
north bridge (memory controller, &c.).  Has anyone else had problems
with a motherboard with a 686B but an AMD or other (e.g. Samsung)
north bridge on it?